Super-resolution is the process of using one or more low resolution images to create an enhanced, higher resolution image; either through heuristics or by image stacking.
What is Single-frame SR?
Initially developed for use in video-game emulators, single-frame super-resolution methods seek to increase the resolution of an image by recognizing shapes and redrawing the same at a higher resolution. These methods work well for low-bit-count computer-generated graphics such as line-drawings and classic game graphics, but generally perform poorly with photographic images. The IES implementation is based on a custom extension of the original ideas. This extension allows for alpha channel processing and 32 bits per pixel input.
What is Heuristic Scaling?
This transform performs heuristic scaling (a type of single-frame super-resolution) on the current layer:
Original
3x Heuristics Super-resolution
The preview image is done in draft
quality, however, when you click OK the whole operation is repeated in the
maximum possible quality. Thus, the final image will usually look better than it did
when you were setting transform parameters.
